How accurate is the US November CPI?

【Editor's Note】After the release of the November US CPI data, which significantly missed expectations, the market was cautiously optimistic—on the one hand, it welcomed the weaker-than-expected inflation data, but on the other hand, it was skeptical of the authenticity of the inflation data due to the government shutdown. This skepticism is reasonable.

I. Three interpolation methods for CPI and how it is handled after a government shutdownBLSUnited States The calculation of CPI relies on the collection of approximately 80,000 price quotes each month, but when data is missing, BLS uses three main interpolation methods to fill in the gaps, ensuring the continuity of the index as much as possible. These methods are applied in order of priority: First is Cell-relative Imputation (BLS), which uses the average price change of similar items in the same geographic area and product category to estimate missing values. For example, if the price of milk in a supermarket could not be collected for some reason, BLS will refer to the average price change of milk in other stores in the same area to estimate the missing data; second is Class-mean Imputation. The text="">) is filled by changes in more generalized similar commodities; finally, there is "Carry Forward", which directly copies the prices from the previous month, assuming no change.

In2025year10month1Sunday11BLSTake a lot ofCarry Forward, translate the monthly data from 9 directly to 10month, as

Then, BLSwill multiply the 7% to10

The result is that rentCPIis 9-11Month's compound growth rate has changed from what it should have12) is undervalued to 7% (from 10To10.7). PS: The reason why we don't directly use 12 as the rent price index for November is because, in reality, BLS obtains several thousand sample rents instead of a single rent price—it cannot directly calculate the price index 12; it can only first calculate the weighted growth rate and then multiply it by the base period price. In summary, for ordinary goods and non-real estate services, even if the October data is artificially suppressed due to carry-forward, as long as the November data itself is accurate, then... The compound inflation rate for September to November is relatively accurate. However, the housing component cannot achieve this self-correction. 10Once the monthly data iscarry forward, then 11monthly housingCPIIt is bound to be underestimated. The underestimation is precisely the housing inflation in October itself. This is why, compared to previous months, the decline in rent inflation this month is significantly greater than that of core goods and non-housing services (Figure 1), while the year-on-year CPI growth rate excluding housing has not changed much from the previous two months (Figure 2). Based on this estimate, the 0.18% compound annual growth rate of the housing sub-item this month is actually the single-month growth rate of October-November. This also aligns better with the month-on-month growth rate of housing prices from June to September (approximately 0.26%). Figure 1: Inflation of Housing, Core Goods and Non-Housing Services Figure 2: Annual CPI Growth Rate Excluding Housing

Data Source: Haver, GMF Research

III. Other Factors That May Depress Other Inflation Components

Of course, we do not believe that other components are necessarily accurate.

After all, the data collection period in November was shorter than in other months. We speculate that three other factors contributed to the underestimation of the non-housing component, but the underestimation may not have been as significant as that of the housing component. First, there is the lag in weight adjustments. Normally, BLS adjusts component weights based on relative price changes, typically increasing the weight of high-inflation components. BLS explicitly states that the lack of data for October resulted in no weighting adjustment, which may have slightly underestimated overall inflation. Secondly, there is the issue of the data collection time window. Because data collection for November was delayed by about two weeks compared to usual, some collection work may have coincided with the holiday promotional season, such as Black Friday. This could cause the seasonal adjustment factor, which was originally used to eliminate seasonal fluctuations for the entire month of November, to become ineffective. Finally, it cannot be ruled out that some samples in the data collection for November itself were not recorded in time and continued to use previous values, but it is not clearly stated how much data analysis comes from September. Carry Forward Carry Forward。

IV. Overall Commentary on Inflation Data and Monetary Policy

Since Carry Forward primarily affects housing, the super core inflation excluding housing and used cars may still be of reference value.Carry Forward

IV. Overall Commentary on Inflation Data and Monetary Policy

Since Carry Forward primarily affects housing, the super core inflation excluding housing and used cars may still be of reference value.Carry ForwardCarry Forward